[15] RFR: 8202117: com/sun/jndi/ldap/RemoveNamingListenerTest.java fails intermittently: Connection reset

Chris Yin xu.y.yin at oracle.com
Mon Mar 16 07:26:41 UTC 2020


Thanks for reviewing, Daniel, Vyom.


Hi, Daniel

I modified the test as you suggested to cover the potential issue with URIBuilder, many thanks. Updated webrev as below:

http://cr.openjdk.java.net/~xyin/8202117/webrev.01/

Regards,
Chris


> On 13 Mar 2020, at 7:51 PM, Daniel Fuchs <daniel.fuchs at oracle.com> wrote:
> 
> Hi Chris,
> 
> This looks fine to me too. Thanks for taking care of this issue.
> 
> A potential issue I see is that the test might fail if
> "localhost" does not resolve to the loopback address - but you
> would likely get a "Connection refused" in that case.
> 
> One possibility to get that out of the way could be to use
> the URIBuilder:
> 
> * @library lib/ /test/lib
> 
>   ...
> 
>           URI providerURI = URIBuilder.newBuilder()
>                .scheme("ldap")
>                .loopback()
>                .port(server.getLocalPort())
>                .path("/o=example")
>                .build();
>   ...
> 
>  59         env.put(Context.PROVIDER_URL, providerURI.toString());
> 
> best regards,
> 
> -- daniel
> 
> 
> On 13/03/2020 08:28, Chris Yin wrote:
>> Hello,
>> Please review following changes to try to fix intermittent failure of test com/sun/jndi/ldap/RemoveNamingListenerTest.java, thanks
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8202117
>> Webrev: http://cr.openjdk.java.net/~xyin/8202117/webrev.00/
>> According to failure logs, test already run done and give a pass message, but test framework caught “java.lang.RuntimeException: java.net.SocketException: Connection reset” from other thread during test end, go through the test code, LDAPServerHandler thread may throw such exception in specific case. This change will replace test itself implemented TestLDAPServer/LDAPServerHandler with customized BaseLdapServer to fix the corner. I had run the changed test on 4 platforms for total 600 times, no failure observed.
>> Thanks,
>> Chris
> 



More information about the core-libs-dev mailing list